Skip to content

CLI Reference (VE_CLI.exe)

VE_CLI.exe lives alongside Visual Enhancer.exe and runs every major image and video workflow headlessly (QT_QPA_PLATFORM=offscreen), reading presets exported from the desktop app and emitting structured progress or machine-readable JSON.

Terminal window
VE_CLI.exe <COMMAND> [INPUT ...] [OPTIONS]
VE_CLI.exe <COMMAND> --help

INPUT may be one or more individual files or folders (pass -- before filenames starting with -).

Command Purpose Default Active Stages (without --preset)
render Full Neural Rendering and/or ReShade pipeline neural_model, or reshade (if only --reshade-* flags are passed), or both
reconstruct Native-resolution (1x) DLSS Ray Reconstruction dlss_ray_reconstruction
reshade Apply a .ini ReShade preset to images or videos reshade
upscale Upscale images or videos using RTX VSR, DLSS, or AI models Selected engine (rtx, dlss, visionnext, razornext, peakmagnetar, optipulsar)
interpolate Video frame generation via DLSS or AMD FSR 3 frame_generation (DLSS) or fsr_frame_generation (FSR 3)
devices Enumerate NVIDIA CUDA GPUs, ReShade status, and DXGI FSR 3 adapters Does not accept input files
Flag Values / Format Description
-o, --output PATH Directory or filename Output folder, or explicit output filename when given one input whose extension matches image/video rules. Default: <app-folder>\outputs\
--preset JSON File path Load an exported app .json preset; CLI flags override matching fields
--recursive Switch Recurse into subfolders when an input path is a directory
--gpu ID auto, NVIDIA index (0), or GPU UUID Select NVIDIA GPU for CUDA/NGX/ReShade stages (fails if GPU is non-RTX when RTX is required)
--codec VALUE h264, hevc, av1, vp9, prores, ffv1 Video encoder selection (prores selects ProRes HQ). For NVENC hardware encoders or ProRes Proxy, configure in a --preset JSON
--container VALUE mp4, mkv, mov, webm Video mux container (must be compatible with codec; specifying --container webm without --codec selects VP9)
--quality VALUE auto, good, best, max Constant encoding quality tier
--dry-run Switch Print resolved pipeline stages, dimensions, FPS, and HDR state without initializing hardware or rendering
--quiet Switch Suppress live progress bar output
--json Switch Print a single compact JSON summary on stdout (progress and errors go to stderr)

Image Output Options (render, reconstruct, reshade, upscale)

Section titled “Image Output Options (render, reconstruct, reshade, upscale)”
Flag Values / Format Description
--format VALUE png, jpeg, webp, avif, tiff, jxl Image output format (--format cannot be used with video inputs)
--image-quality N 1..100 Encoder quality for JPEG, WebP, AVIF, and JPEG XL (100 enables lossless JPEG XL)
--bit-depth N 8 | 16 Output precision (16 supported on png, tiff, and jxl)
--jxl-lossless N 0 | 1 Pixel-lossless JPEG XL compression
--jxl-effort N 1..10 JPEG XL compression effort (default 7; higher is slower)
--jxl-progressive N 0 | 1 Enable progressive JPEG XL decoding
Flag Values / Format Description
--style N 0 | 1 | 2 Neural Rendering style (Style 0, Style 1, Style 2). Errors if --preset disables neural_model
--strength X 0..2 Neural Rendering intensity multiplier (0.0 to 2.0)
--reshade-* See reshade below Any --reshade-* flag works on both render and reshade
Flag Values / Format Description
--flow-quality VALUE high | medium | low Video NVOFA optical-flow quality for DLSS Ray Reconstruction
Flag Values / Format Description
--reshade-preset PATH .ini file path ReShade preset file
--reshade-shaders PATH Folder(s), ;-separated Search paths for .fx / .fxh shaders
--reshade-textures PATH Folder(s), ;-separated Search paths for ReShade textures and LUTs
--reshade-mix X 0..1 (or 0..100 integer %) Blend factor between source (0) and ReShade output (1 / 100%)
--reshade-depth VALUE none | auto | moge-2-vitb Depth source (none/off, auto, moge-2-vitb/ai)
--reshade-temporal VALUE auto | quality | off Video temporal optical-flow stability (amd-optical-flow alias maps to quality)
Flag Values / Format Description
--engine VALUE rtx, dlss, visionnext, razornext, peakmagnetar, optipulsar Select upscaling engine/model (ai also selects AI Upscale)
--model VALUE visionnext, razornext, peakmagnetar, optipulsar Explicitly select the AI upscale model when using an AI engine
--scale FACTOR Image AI: 1, 2, 4, 6, 8; Video AI: 2, 4; RTX: >= 1 Output scale multiplier (mutually exclusive with --size and disallowed with --engine dlss)
--size WIDTHxHEIGHT 2..32768 per side Exact RTX target size or supported AI Video Target Resolution (640x480, 1280x720, 1920x1080, 2560x1440, 3840x2160, 7680x4320)
--mode VALUE dlaa, quality, balanced, performance, ultra-performance DLSS Super Resolution preset (--engine dlss only)
--memory-budget VALUE auto, unlimited, 12, 8, 6 RazorNext GPU VRAM budget in GB (--engine razornext only)
--tile VALUE auto, 256, 512 VisionNext tile size (razornext accepts only auto; Magnetar manages tiles automatically)
--upscale-gpu ID / --visionnext-gpu ID auto or DXGI ID (dxgi:...) Adapter selection for AI upscale models (--visionnext-gpu is an alias for --upscale-gpu)
Flag Values / Format Description
--interpolator VALUE dlss | fsr3 Frame-generation provider (default: dlss)
--fps FPS e.g. 60, 120, 59.94 Target output frame rate (mutually exclusive with --multiplier)
--multiplier N 2..8 Multiply source frame rate while preserving rational fractional cadence
--fsr-gpu ID auto, DXGI index, or DXGI ID Adapter selection when --interpolator fsr3 is set
Exit Code Meaning
0 Success (all files rendered, devices completed, --dry-run planned, or --help printed)
1 Runtime error, decode/encode failure, missing model, or at least one file failed in a batch
2 Usage error (unknown command/flag, conflicting options, invalid preset, or failed settings validation)
130 Cancelled by user (Ctrl+C, Ctrl+Break, or console close event)
Terminal window
$hw = .\VE_CLI.exe devices --json | ConvertFrom-Json
$hw.devices | Format-Table index, name, memory_mb, ai_compatible, reshade_available
$hw.fsr_adapters | Format-Table dxgi_index, name, available, id

2. Dry-Run a Recursive Batch Before Rendering

Section titled “2. Dry-Run a Recursive Batch Before Rendering”
Terminal window
.\VE_CLI.exe render .\captures --recursive --preset .\cinematic.json --dry-run
Terminal window
.\VE_CLI.exe upscale .\screenshots `
--engine razornext `
--scale 4 `
--memory-budget 8 `
--format jxl `
--bit-depth 16 `
--jxl-lossless 1 `
-o .\upscaled_jxl

4. Apply a ReShade Preset to a Video and Double Frame Rate via Preset

Section titled “4. Apply a ReShade Preset to a Video and Double Frame Rate via Preset”
Terminal window
# Export a preset with ReShade + FSR 3 / DLSS Frame Generation active, then override paths per folder:
.\VE_CLI.exe render .\gameplay.mp4 `
--preset .\reshade_fg_preset.json `
--reshade-preset "C:\ReShade\Filmic.ini" `
--reshade-shaders "C:\reshade-shaders\Shaders" `
--reshade-textures "C:\reshade-shaders\Textures" `
--codec av1 `
--container mkv `
--quality best `
-o .\rendered\gameplay_enhanced.mkv